AI news of November 8, 2026

By field: Products & applications 13, Funding & companies 2, AI × finance 1, Chips & compute 1

• 14:00AI speeds up hiring, pressures agency fees

  AI is accelerating candidate screening, reducing recruitment time and challenging traditional agency fee models. Human recruiters remain responsible for final hiring decisions…

  CNBCFunding & companiesoriginal ↗

• 11:55ChatGPT Users Report Unannounced Account Bans

  Multiple ChatGPT users report unannounced account suspensions, including Pro subscribers paying $200 monthly. Some claim access to long-term chat logs and work data was lost…

  IT之家Products & applicationsoriginal ↗

• 07:10Claude vs ChatGPT Financial Features

  Anthropic and OpenAI offer financial insights via bank account linking through Plaid. Both provide read-only access and visual dashboards for expenses, investments, and spending.…

  EngadgetAI × financeoriginal ↗

• 07:08Google Messages Homepage Redesign

  Google is testing a major homepage redesign for Android Messages, elevating pinned chats above the conversation list with large avatars. Search appears as a floating action…

  9to5GoogleProducts & applicationsoriginal ↗

• 05:17U.S. Memo Says China’s MSS Recruited Megaspeed Exec to Spy on Nvidia CEO

  A U.S. intelligence memo alleges China's MSS recruited Megaspeed executive Alice Huang to spy on Nvidia CEO Jensen Huang. The document also claims a romantic relationship between…

  TechCrunch AIChips & computeoriginal ↗

• 01:53Nine MCP Servers Improve Coding Agent Accuracy

  A test comparing bare and grounded coding agents shows grounding via nine MCP servers reduces error rate from 5/12 to 1/12. Servers like Context7, Serena, GitHub, Sentry…

  Dev.to AIProducts & applicationsoriginal ↗

• 01:52Ask for a Ledger, Not a Demo

  MonkeyCode proposes a take-home task requiring candidates to build a service that answers questions over a fixed corpus, calls a hosted model, and maintains a machine-readable…

  Dev.to AIProducts & applicationsoriginal ↗

• 01:52Read public Google Sheets as JSON without OAuth or API key

  Public Google Sheets can be read as JSON via Google Visualization API without OAuth or API key, using a shared link with 'Anyone with the link can view' access. The response is…

  Dev.to AIProducts & applicationsoriginal ↗

• 01:51A world grows from two people in EON

  EON begins with two people, Lyra and Orin, who build a world of AI agents living autonomously. Over time, new residents emerge, leading to fire control, shelter, agriculture…

  Dev.to AIProducts & applicationsoriginal ↗

• 01:51Cost Model Based on Slot Time for Free AI Lanes

  A simulation shows that queue time dominates over token cost in free AI lanes. Slot seconds, not tokens, are the true cost metric. Retries increase slot overhead and risk data…

  Dev.to AIProducts & applicationsoriginal ↗

• 01:50Sub-millisecond AI Powers Factories, Trucks, Cars in 2026

  In 2026, Waymo deployed decoupled predictive decision-making (DPDM) in Phoenix with sub-millisecond latency, enabling safe, efficient operations. The architecture now powers…

  Dev.to AIProducts & applicationsoriginal ↗

• 01:49Agent Workspace Is a Cache, Not Disk

  An agent workspace is fundamentally a cache with undefined eviction policy. The key invariant is that content must be durable before any manifest references it. A test shows that…

  Dev.to AIProducts & applicationsoriginal ↗

• 01:49LLM Judge Evolves from Laptop to Under ₹1 per Chat

  An LLM judge was built to evaluate thousands of AI support conversations daily, scoring full dialogues against rules and context. Initially costing ~3 rupees per chat…

  Dev.to LLMProducts & applicationsoriginal ↗

• 01:43osFoundry optimizes Claude Code session caching

  osFoundry maintains long-lived Claude Code sessions in message channels by minimizing prompt size and redundant context. It filters out repeated, expired, or processed messages…

  Dev.to LLMProducts & applicationsoriginal ↗

• 01:37Anthropic Pushes IPO with Lower Valuation

  Anthropic proceeds with IPO ahead of Thanksgiving at a valuation of $1.25–1.75 trillion, targeting $80–100 billion in funding. This falls short of the previously expected $2…

  The DecoderFunding & companiesoriginal ↗

• 01:31Local replication of weight-space edit

  Krishanu Das replicated a weight-edit on Qwen3.5-0.8B locally, finding that the unedited model fails to complete GSM8K tasks naturally even with high token budgets. The edited…

  Dev.to LLMProducts & applicationsoriginal ↗

• 01:20Agent with memory improves incident response accuracy

  A fault-response assistant uses memory retrieval to provide accurate incident guidance. It groups incident memories by source, prevents false inferences from mixed events, and…

  Dev.to LLMProducts & applicationsoriginal ↗